Hacker News Reader

Comment by strict9 | Hacker News Reader

This issue isn't what need solving. The much larger problem is the "unknown or expired link" page.

What year is this? Why are we still accepting an implementation detail as an excuse for an awful user experience?

Yeah this is the biggest problem with HN.

Go to front page, click an article or 2 and click to go to next page and its already "expired" in minutes. For a site that caters to startups/developers its pretty embarrassing.

Yes. I hit that page almost every time I hit the 'more' link.

What I resort to doing on bursts of high HN reading (such as now, at lunch) is to command+click each link that looks mildly interesting and go through a few pages.

This site is too frustrating to use any other way, and for a community with so many engineers, I'm still shocked that so many (including myself) put up with it.

The only reason any of us put up with it is the community.

It doesn't matter how many engineers there are here, it's Paul Graham's forum and he seems unwilling to fix these sorts of issues. There have been several threads where I've seen offers to fix it.

HN still uses tables for designs. There's a mix of CSS and <font> tags. The login/register page has no styling at-all. Searching is handled by a third-party (HNSearch). Your top-bar color won't persist on some pages (ie. submit).

HN has a lot of issues but yes, by far the most annoying is the link expiration.

all the ones you listed are trivial though.

tables still > css

Well what's the solution?

Do you simply redirect all "expired link" GETs to the home page? Should we have a fixed # of 10 first pages (like some imageboards)?

Maybe it should redirect to the New page.

The solution is to avoid capturing the state of the user on the server, passing it in the URL instead. That way nothing need expire about the link.

The solution is for someone else to create a better site that's actually maintained and start migrating people away from HN. A more neutral foundation would be better anyway, imo.

User state is current captured in a closure in the single process single threaded HN server (running a custom lisp). The solution is to not do that.

Reply on news.ycombinator.com